Commits · 828f862acb5f46ffaa1633aa80d85af73c31c97a · OpenDAS / vllm_cscc · GitLab

18 Mar, 2026 40 commits
- [Bugfix] Expand quantization method support in perf metrics (#37231) · 828f862a
  Thillai Chithambaram authored Mar 18, 2026
```
Signed-off-by: Thillai Chithambaram <thillaichithambaram.a@gmail.com>
```
  828f862a
- [Bugfix] Fix KV scales inconsistency in fp8 MLA & FlashInfer kv_cache_dtype... · 577df69b
  Andy Lo authored Mar 18, 2026
```
[Bugfix] Fix KV scales inconsistency in fp8 MLA & FlashInfer kv_cache_dtype "auto" leading to gibberish (#37054)
Signed-off-by: Andy Lo <andy@mistral.ai>
```
  577df69b
- [Model Runner V2] Spec decode rejection sampler greedy support (#37238) · 04244fd0
  Giancarlo Delfin authored Mar 18, 2026
```
Signed-off-by: Giancarlo Delfin <gdelfin@inferact.ai>
```
  04244fd0
- [Bugfix] Remove assertion for NVFP4 scale dynamic range (#37465) · 9482b0b0
  Michael Goin authored Mar 18, 2026
```
Signed-off-by: Michael Goin <mgoin64@gmail.com>
```
  9482b0b0
- [LoRA][BugFix] Fix skipped LoRA adapters for Mistral3 (#36928) · 5bc1da14
  Woosuk Kwon authored Mar 18, 2026
```
Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>
```
  5bc1da14
- fix(worker): optimize swap_states to copy only active token prefixes (#34733) · 00910171
  Philip Ottesen authored Mar 18, 2026
```
Signed-off-by: Philip Ottesen <phiott256@gmail.com>
```
  00910171
- [V0 Deprecation] Deprecate virtual engine (#37195) · 0d81a1fe
  Wentao Ye authored Mar 18, 2026
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  0d81a1fe
- chunk parakeet into 30s clips to prevent OOMs on long audios (#36671) · 6ae4c8d6
  Netanel Haber authored Mar 18, 2026
```
Signed-off-by: Netanel Haber <58652339+netanel-haber@users.noreply.github.com>
```
  6ae4c8d6
- [Bugfix] Fix ROCm crash in qwen3_next multi-stream events (#36795) (#37427) · a913b612
  JartX authored Mar 18, 2026
```
Signed-off-by: JartX <sagformas@epdcenter.es>
```
  a913b612
- Fix models which use `layer_type_validation` for Transformers v5 (#37398) · 5ce2d10e
  Harry Mellor authored Mar 18, 2026
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  5ce2d10e
- [Bugfix] Fix incorrect use of merge_size in Qwen3-VL video timestamp calculation (#37439) · 738d0a28
  Chengyu Fang authored Mar 19, 2026
```
Signed-off-by: chengyufang <cnyvfang@outlook.com>
```
  738d0a28
- [bugfix][async scheduling] fix extra cuda context in device 0 with EP/DP (#37449) · 70b81c4f
  youkaichao authored Mar 19, 2026
```
Signed-off-by: youkaichao <youkaichao@gmail.com>
```
  70b81c4f
- [Model] Remove unnecessary processor definition for Nemotron Parse (#37456) · 7476d148
  Cyrus Leung authored Mar 19, 2026
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  7476d148
- [Misc] Clean up model registry (#37457) · f3732bd9
  Cyrus Leung authored Mar 19, 2026
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  f3732bd9
- [Perf] Add tuned triton moe config for Qwen3.5 H200, 9.9% E2E throughput improvement (#37340) · 0ef7f790
  Wentao Ye authored Mar 18, 2026
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  0ef7f790
- [kv_offload+HMA][2/N]: Support multiple KV groups in GPULoadStoreSpec (#36642) · 5dd8df07
  Or Ozeri authored Mar 18, 2026
```
Signed-off-by: Or Ozeri <oro@il.ibm.com>
```
  5dd8df07
- Add API docs link if the CLI arg is a config class (#37432) · 39bfb57b
  Harry Mellor authored Mar 18, 2026
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  39bfb57b
- Adding deterministic lora benchmarking to vLLM Bench (#36057) · c9d838fc
  RonaldBXu authored Mar 18, 2026
```
Signed-off-by: Ubuntu <ubuntu@ip-172-31-43-201.ap-northeast-1.compute.internal>
Signed-off-by: Ronald Xu <ronaldxu@amazon.com>
```
  c9d838fc
- [Kernel] Add gpt-oss Router GEMM kernel (#37205) · b1169d7b
  Xin Yang authored Mar 18, 2026
```
Signed-off-by: Xin Yang <xyangx@amazon.com>
```
  b1169d7b
- standardize load_weights using AutoWeightsLoader for kimi_linear and minimax_text_01 (#37371) · 17808394
  XLiu-2000 authored Mar 18, 2026
```
Signed-off-by: XuLiu <xuliu40@gmail.com>
Co-authored-by: XuLiu <xuliu40@gmail.com>
```
  17808394
- [Perf] Eliminate padding and slicing op for GPT-OSS with Flashinfer MXFP4 MXFP8 MoE (#30647) · 296839a1
  elvischenv authored Mar 18, 2026
```
Signed-off-by: elvischenv <219235043+elvischenv@users.noreply.github.com>
```
  296839a1
- [Log] Reduce duplicate log (#37313) · c373b5c0
  Wentao Ye authored Mar 18, 2026
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  c373b5c0
- elastic_ep: Fix stateless group port races (#36330) · de1a86b7
  Itay Alroy authored Mar 18, 2026
```
Signed-off-by: Itay Alroy <ialroy@nvidia.com>
```
  de1a86b7
- [2/3] Refactor InternVL-based processors (#37324) · 99267c23
  Cyrus Leung authored Mar 18, 2026
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  99267c23
- [kv_offload+HMA][6/N]: Split offloading_connector.py (#37405) · 525f2eeb
  Or Ozeri authored Mar 18, 2026
```
Signed-off-by: Or Ozeri <oro@il.ibm.com>
```
  525f2eeb
- [Bugfix] Fix base64 JPEG video frames returning empty metadata (#37301) · 918b7890
  Yufeng He authored Mar 18, 2026
```
Signed-off-by: Yufeng He <40085740+universeplayer@users.noreply.github.com>
Signed-off-by: Yufeng He <40085740+he-yufeng@users.noreply.github.com>
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Co-authored-by: Yufeng He <40085740+universeplayer@users.noreply.github.com>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
  918b7890
- [NIXL][Bugfix] metrics & testing minor bug (#36051) · 98b09ddc
  Andy Lo authored Mar 18, 2026
```
Signed-off-by: Andy Lo <andy@mistral.ai>
```
  98b09ddc
- [Model] Enable LoRA support for tower and connector in H2OVL (#31696) · cef1f302
  Shwetha Poojary authored Mar 18, 2026
```
Signed-off-by: shwetha-s-poojary <shwetha.s-poojary@ibm.com>
```
  cef1f302
- [Bugfix] Fix EP weight filter breaking EPLB and NVFP4 accuracy (#37322) · 17c47fb8
  Elvir Crnčević authored Mar 18, 2026
```
Signed-off-by: Elvir Crncevic <elvircrn@gmail.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Kevin H. Luu <khluu000@gmail.com>
```
  17c47fb8
- [Build] Bump python openai version (#32316) · b322b197
  Chauncey authored Mar 18, 2026
```
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
```
  b322b197
- [CI] Fix PaddleOCR-VL HF test failure due to create_causal_mask API rename (#37328) · eaf7c9b9
  Andreas Karatzas authored Mar 18, 2026
```
Signed-off-by: Andreas Karatzas <akaratza@amd.com>
```
  eaf7c9b9
- [docs] Add docs for new RL flows (#36188) · 47a1f11b
  Aaron Hao authored Mar 18, 2026
```
Signed-off-by: ahao-anyscale <ahao@anyscale.com>
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  47a1f11b
- fix(glm47): improve tool call parsing and content normalization (#37386) · fad09e8a
  Karan Bansal authored Mar 18, 2026
```
Signed-off-by: karanb192 <karan@example.com>
Co-authored-by: karanb192 <karan@example.com>
```
  fad09e8a
- [LoRA] Make LoRA respect `language_model_only` (#37375) · 8c31f47c
  Jee Jee Li authored Mar 18, 2026
```
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
```
  8c31f47c
- [Bugfix] Avoid OpenMP thread reallocation in CPU torch compile (#37391) · 26180124
  Li, Jiang authored Mar 18, 2026
```
Signed-off-by: jiang1.li <jiang1.li@intel.com>
```
  26180124
- [kv_offload+HMA][0/N]: Support block-level preemption handling (#34805) · fcf0687b
  Or Ozeri authored Mar 18, 2026
```
Signed-off-by: Or Ozeri <oro@il.ibm.com>
Co-authored-by: Nicolò Lucchesi <nlucches@redhat.com>
```
  fcf0687b
- [XPU] skip unsupported ut and update test_nixl_connector (#37179) · 86b7e3c9
  liuzhenwei authored Mar 18, 2026
```
Signed-off-by: zhenwei-intel <zhenwei.liu@intel.com>
Co-authored-by: Kunshang Ji <kunshang.ji@intel.com>
```
  86b7e3c9
- [responsesAPI] parser.extract_response_outputs can take in token IDs (#37130) · 0e959161
  Andrew Xia authored Mar 17, 2026
```
Signed-off-by: Andrew Xia <axia@meta.com>
```
  0e959161
- [CI] Stabilize test_cpu_offloading by waiting for async offload before cache reset (#37335) · ce2ef42f
  Andreas Karatzas authored Mar 18, 2026
```
Signed-off-by: Andreas Karatzas <akaratza@amd.com>
```
  ce2ef42f
- [ROCm][CI] Add ROCM_EXTRA_ARGS to audio_in_video test server fixture (#37349) · 8b632575
  Andreas Karatzas authored Mar 17, 2026
```
Signed-off-by: Andreas Karatzas <akaratza@amd.com>
```
  8b632575